Buoy benchmark: Claude runs apps 6x faster and cheaper with MCP tools than none
LovesWorkin · reddit · 2026-10-14
Buoy's lead dev ran a controlled benchmark: Claude Sonnet 5.5 in Claude Code drove a food-ordering test app on iOS Simulators, 14 jobs × 3 runs, varying only the toolkit. No tools: 17m44s, ≥$5.92, 31/42 passed; Argent 0.27.0: 6m26s, $5.71, 40/41; Buoy MCP 7.0.63: 2m56s, $2.60, 41/42. Caveats: per-step, no tools was fastest (5.5s vs Buoy 7.1s, Argent 46.4s), and Buoy was tuned on these jobs. Net: toolkits make agents faster, cheaper and more reliable, with real latency trade-offs.
More from coding & agent
- Beancat launches an AI agent for media production with transparent per-run dollar pricing — Scobleizer · 2026-10-14
- Two devs burn $2,025 in API costs building a game with Claude, screenshots show progress — zeeg · 2026-10-14
- Sketch first, code later: prototyping with tldraw before handing specs to Claude — max__drake · 2026-10-14
- Google ran Argon overnight on a 200K-line codebase: 40% faster sessions, 2 of 6 fixes shipped — rseroter · 2026-10-14
- Argon synthesizes its own voice from scratch: pure Python, no audio libraries — fofrAI · 2026-10-14
- Musk RTs: Grok Bot Multi-Agent Team Does Your Grocery Shopping in 3 Minutes — elonmusk · 2026-10-14